Puget Systems print logo
Read this article at https://www.pugetsystems.com/guides/784
Richard Falk (Customer Support)

CPU Coolers and My Experiment

Written on March 30, 2016 by Richard Falk

I wanted to double check the accuracy of a certain CPU cooler from Noctua. Mainly for my own curiosity as well as part of our ongoing process to ensure we are only selling high quality parts to our customers. To your right is a Noctua NH-U12S CPU cooler.  It did very well through our testing, and was added to our product line. When we qualify a new product to be a fit for our product line, everyone is emailed all of the details about the product and given a chance to offer opinions or concerns as part of the process. I had some concerns based on a combination of the pictures of the product and the manufactures specifications not quite lining up in my mind. 

Before I get into it, if you do not know much about CPU coolers, I want to give you a quick run down of how the air coolers work.

Heatsinks use either copper, aluminum, or a combination of the two in order to move heat from the base of the cooler through heatpipes to the heatsink. A fan then blows air through the heatsink to move the heat into the air, and then out of the system, effectively keeping the CPU within safe operating temperatures. In contrast, a liquid cooling system essentially uses the same idea but replace the copper heatpipes with tubes filled with a thermally conductive liquid that is pumped to a radiator. Copper is a very good thermal conductor, aluminum is about half as good, and air is absolutely pathetic. I will explain the differences in further detail in the next section.


How does a CPU cooler work exactly?

Thermal conductivity (how well materials transfer heat) is measured in watts per meter Kelvin. I know that sounds like gibberish, but we do not need to understand the science so much as to pull the nice even numbers scientists have worked hard to get so we can have a real number we can look at to gauge the differences between materials. These numbers here are not absolute, as different purities of materials have different properties, so you will see all sorts of numbers if you do some research on the topic. But for comparative purposes, lets use these numbers I have found.  

  • Copper's thermal conductivity is 385 W/m*K
  • Aluminum's is about 205 W/m*k
  • Arctic silver MX-2 thermal paste (we use this on most systems) 5.6 W/m*K
  • Air's thermal conductivity (measured at 0c) is about 0.024 W/m*K

This is the same reason why the sun takes many hours to heat up the air in the morning, but it takes only a few minutes to heat up a piece of metal (if you have morning frost you will notice this).

So the way a CPU air cooler works is pretty simple!

A CPU cooler consists of its heatsink, heatpipes, and base. In addition there is the heatplate on the CPU and the thermal paste in between the CPU and the CPU cooler.

A base (usually copper, but aluminum on lower cost units) connects directly to the heatplate on the top of a CPU. The heat then gets transferred from the CPU heatplate through the thermal paste, which is used to ensure any gaps between the CPU heatplate and the CPU cooler base are filled. The base then connects to heatpipes in the CPU cooler. The base and heatpipes typically contain copper, as copper's thermal conductivity is very high. 

The heatpipes have a solid shell, a porous inside, and a hollow center that contains a liquid/gas. As heat gets transferred from the base of the cooler to the heatpipe, it heats up the liquid to the point of evaporation which turns it into gas. Once in gas form, it moves through the center of the pipe until it gets cool enough to condense back to a liquid. Once it is liquid again it flows through the porous material until it gets back to the base of the heatsink to get hot enough to evaporate again. 

The heatsink contains a large array of fins, typically made from aluminum.  The heatpipe will deliver heat into the fins, which are spread out thinly so that the airflow through them can dissipate heat easily.  Since they're designed to cover a large area, they don't need the high level of thermal conductivity that the heatpipes and base need.

Now, going back to why I wrote this - the manufacturer specifications said the materials were a copper base and heat pipes with aluminum cooling fins. Copper is the color of a penny and aluminum is the color you see on the Noctua cooler up above. Because they stated that it was copper pipes and a copper base which you would expect on a really high end cooler to have, I was confused when in the pictures of the product, it looked like aluminum. I was skeptical of the manufacturers statement that it was zinc covered copper. So I went and got a heatsink and (with my boss's permission) started to use a filing tool to file off the metal to prove if that was an accurate statement. After removing the coating of zinc, I discovered that indeed, the base and heatpipes were made of copper. Not only were my fears resolved, but Noctua was proven by fire once again to have a solid product. But I was able to have fun with it! Here are some pictures to prove I can break things too! 

Here you can see the heatsink which has a copper heatpipe, and the copper base which were exposed after filing off the zinc coating. Zinc does not have as good of thermal conductivity as copper or aluminum, but being as it is such a small amount of it (a very thin layer like paint) It does not really affect the cooling ability of the unit. It was likely done as Zinc is resistant to corrosion and copper as we know can corrode.

After reading this, you probably think I like to break some of our new products, but that is not the case at all. And to make you feel better about me breaking perfectly nice things, after I was done attacking this cooler, I sanded down and cleaned the unit and we will actually use it internally for a specific use. After reading this article, I hope you have a better understanding of how CPU coolers work and also learned a little bit about how we test new products.

Tags: Cooling, Noctua, CPU, heatsink, heatpipe, copper, aluminum, puget, systems
Braden Snyder (lishde)

good read

Posted on 2016-06-21 00:44:14

Not zinc coating - nickel. Nickel is much harder and less reactive than zinc. Galvanized steel has a zinc coating, which is silvery, but not shiny like nickel

Posted on 2016-08-01 22:51:12
Mike Halterman

Very cool. I was impressed with both your diligence and your company's willingness to spend some extra cash to make sure the components you use are the best. Not only did I learn something, but I am even more committed to using Puget for my next computer.

Posted on 2019-04-24 20:09:43

NH-U12S ~vs~ DH-D15
The amount of data and research Puget Systems provides is truly mind blowing.
You include the Noctua NH-U12S in a lot of your base build specs.
Can you explain why you have not mentioned the Nocuta NH-D15 ?
It just looks like an obvious contender for upgrading from the NH-U12S.

Cooler Temp 4.5GHz Temp 4.8GHz Noise Level
D15 65.5C 76.8C 47.9 dB(A)
U12S 70.8C 84.6C 45.8 dB(A)

courtesy: Florian Maislinger, pcbuildersclub.com test results

Posted on 2019-12-04 12:40:03

Size and weight: the D15 weighs 1320g versus the 755g of the U12S, which is a huge difference. When shipping systems around the US, that extra mass puts a lot more pressure on the socket and mounting... and if a heatsink comes loose, it does tremendous damage to the other components in the system (and requires a return to the factory for repair / replacement even if there is no other damage). So if the U12S cools sufficiently, why use a bigger heatsink with increased risk of shipping damage? :)

I think the D15 may also be too big to fit in some of our cases, and its size would overhang RAM slots and thus make swapping memory modules harder as well.

Posted on 2019-12-04 17:29:28

First of all, I do not take for granted that you folks actually read and answer inquiries, so, thank you very much, this is greatly appreciated.
Just briefly, then;
My circumstances do not include those you've mentioned.

Price/Performance is a factor.
Let me run this past you, the added cost is negligible, but the added cooling?
Supposedly, with greater capacity for cooling the CPU, resulting lower temperatures in the CPU:
a) speeds being higher ?

b) speeds sustained longer ?

c) longer CPU life ?

How accurate are these statements (a through c)

Thanks !

Posted on 2019-12-09 23:11:55

If you are not intentionally overclocking (manually setting the system to run at speeds beyond the official rating of the CPU) then as long as there is sufficient cooling to keep all of the components from getting too hot there is no real benefit to having *extra* cooling capacity on top of that.

To put it another way, focusing on just the CPU here, as long as the cooler is able to keep it below its thermal throttling point - even under full load of all cores - that is all you need. Additional cooling beyond that won't make it run faster (unless you overclock, which I do not recommend). If the cooling can't keep it, however, then the CPU will throttle down as it gets too hot. So you don't want to use a cooler that isn't fully capable of handling the CPU's heat under load, or else you will start to get the issue you described as (b).

As for the impact of cooling on CPU life, that I don't have any data on. CPUs in general have a very low failure rate, and we don't have any way (that I am aware of) to determine if the reason a CPU dies is because of excess heat or any other factor.

One thing you didn't mention, however, is noise level. The fan(s) on a CPU cooler - and potentially the pump as well, in the case of liquid cooling - can add to the overall sound made by a computer... so having one that is quieter may be something to consider if that is a factor to you. That is one of the reasons I like the Noctua heatsinks and fans we have been using here at Puget Systems (and which I now use in most of my home computers).

Posted on 2019-12-10 17:27:38

What if I want to overclock the CPU (in my case which is core i9 10900k)? Do you recommend this air cooler? If I want to overclock, which cooler should I choose which have a good performance and a long life (without any fails)?

Posted on 2020-09-27 16:53:54

I have not played around with overclocking in a long time, so I am not in a position to offer specific advice about the best products for that, but in general you want to maximize the amount of cooling in that situation. A lot of folks opt for liquid-cooling, since it can potentially get heat away from the CPU faster and dissipate more heat overall - at the cost of more money, more complex installation, and often more noise (at least compared to a Noctua). There is also the added potential for a leak with such coolers, though with sealed "all-in-one" (AIO) liquid cooling that is minimized. If you did want to use an air-cooler like one of these Noctua models for overclocking, I would encourage adding a second fan on the other side of the heatsink in a push-pull configuration (the fan on the front pulls air in, the fan on the back pushes it out). The larger the heatsink itself and the larger & faster the fans the more overall cooling you will get. Hopefully that info helps you make an informed decision :)

Posted on 2020-09-28 16:22:24